¶ 50. (concurring in part and dissenting in part). Is death a condition precedent to a wrongful death claim? It seems pretty obvious that the answer should be "yes."
¶ 51. Yet, the majority answers this question "no." Acknowledging some displeasure with its own response, the majority attempts to blame the legislature for the majority's interpretation. See majority op., ¶ 46, ¶ 47 n.ll. The legislature could not have intended such a result.
¶ 52. I instead conclude that a wrongful death claim accrues upon death — not before death — and therefore join Justice Crooks in dissenting. As he aptly discusses, it is impossible to read the language of Wis. Stat. § 895.03 or our prior cases "in any way other than that death is a condition precedent, which must be met, before there can be such a lawsuit for wrongful death." J. Crooks' dissent/concurrence, ¶ 93.
¶ 53. I write separately, however, to address two additional infirmities of the majority opinion. First, although I agree with the majority that Robert's estate's claim for medical malpractice is barred by the statute of limitations, I do not join its analysis. In determining that Robert suffered an injury on the date that an *585"infection-producing sponge" was left in his abdomen, the majority's discussion fails to elucidate a clear holding and in fact confuses the law.
¶ 54. Second, I write separately to comment on the majority's discussion of dicta. Instead of providing a clear definition of dicta, the majority provides multiple incomplete definitions. It fails to acknowledge as we have previously explained that Wisconsin has two lines of cases defining dicta, and the majority does not even mention one of them. Rather, with minimal analysis, it dismisses as dicta a difficult proposition from a previous case, avoiding any meaningful discussion. Such an approach fosters an end run around stare decisis and undermines our common law tradition of fidelity to precedent.
I
¶ 55. The circuit court correctly observed that under Wisconsin law, the cause of action accrues on the date of injury and injury is defined as a "physical injurious change." See Fojut v. Stafl, 212 Wis. 2d 827, 831, 569 N.W.2d 737 (Ct. App. 1997). The court explained that in this case, there were three events which were advanced as the possible dates of a physical injurious change to the plaintiffs body: (1) the date "when the sponge was left inside him," (2) the undetermined date, sometime "shortly after the first surgery and prior to the second, when Mr. Genrich developed infection and was running a fever," and (3) the date of the second surgery which "involved an invasive procedure of Mr. Genrich being cut open to remove the sponge . . . ."
¶ 56. Ultimately, the court made a finding that the physical injurious change in this case occurred on the date of the second surgery: "[T]he Court finds that *586the latest point where the injury occurs that forms the first justification for a claim would be the second surgery, which would not have been necessary but for the negligence of the medical staff."
¶ 57. The majority, however, determines otherwise. It states, "It was the negligence during the first surgery that resulted in an infection-producing sponge being present in Robert's abdomen." Majority op., ¶ 19. Further, "the presence of an infection-producing sponge in Robert's abdominal cavity is the type of 'physical-injurious change' discussed in [our case law.]" Id., ¶ 20. The majority concludes: "When the doctors negligently left a sponge inside of Robert, which caused the sepsis that resulted in his death, he sustained an 'injury' that triggered Wis. Stat. § 893.55(lm)(a)'s three-year limitations period." Id.
¶ 58. In focusing on the first surgery, it is unclear whether the majority intends to be making a finding of fact or a conclusion of law. To the extent that the majority is finding the fact of when the physical injurious change occurred, the majority ignores the well-established principle that an appellate court will not disturb a finding of fact of the circuit court unless it is clearly erroneous. The majority offers no analysis as to why it can supplant its finding for that of the circuit court.
¶ 59. If the majority's determination is a conclusion of law, then it fails to elucidate a clear test and confuses the law. I cannot join the majority opinion because I do not understand what the majority holds.
¶ 60. Under the majority's test, when is there a "physical injurious change"? That is, when does the cause of action accrue?
*587¶ 61. Does it accrue on the day that a foreign object is left in the body? I am unsure. The majority's test seems to require more.
¶ 62. Does the majority's conclusion mean that when there is no subsequent physical change, sponges and other foreign objects can be left in a patient's body with impunity? I don't know. But if so, how could the law countenance such a conclusion?
¶ 63. Throughout the opinion, the majority repeatedly refers to the sponge as "infection-producing." See majority op., ¶¶ 18,19, 20, 20 n.7. This grammatical construction suggests that it is not the leaving of a foreign object in the body that is the injury — rather, an injury occurs when a foreign object is left and the object produces an infection.
¶ 64. This begs the question. The majority finds that "the infection-causing sponge was present in Robert's abdomen" on the date of the first surgery. Majority op., ¶ 20 n.7. The majority states further that there is no dispute that "the infection-causing sponge first was present in Robert's abdomen" on the date of the first surgery. Id.
¶ 65. The majority's assertion is only half correct. Certainly, there is no dispute that the sponge was present on the date of the first surgery. Yet, there are no facts in the record that indicate that the sponge was "infection-producing" when the first surgery was performed.
¶ 66. There is nothing in the record that indicates whether the doctors selected a clean, sterile sponge to use during surgery, or whether the sponge was infected when it was initially left in Robert's abdomen. Arguably, *588the sponge started producing an infection at some point after the surgery occurred.1
¶ 67. The majority can point to no facts in the record that demonstrate that the sponge was "infection-producing" the moment it was initially left in Robert's abdomen.2 The majority's unfounded factual assumption that the events occurred simultaneously allows it to evade a more thorough examination of its new test.
¶. 68. Consider the facts of a California case where the events were not simultaneous.3 A curved surgical needle was left in the soft tissue of the patient's *589abdomen. About two and one-half years after the surgery, the patient sought medical treatment because he had enlarged lymph glands and was experiencing attacks of high fever. X-rays revealed the presence of the needle. He was advised by the original surgeon that it was not necessary to remove the needle because it was safely encased in scar tissue. Another surgeon, however, performed the surgery to remove the needle and later testified that "it was necessary to remove the needle, that it had punctured the colon, caused peritonitis and the illness which plaintiff suffered preceding the finding of the needle." Bowers v. Olch, 260 P.2d 997, 1001 (Cal. App. 2d 1953).
¶ 69. If the majority's test was applied to the facts of that case, would the cause of action accrue on the date that the surgeon left the needle which caused the subsequent rupture, peritonitis, and fever? I think so, but am unsure. It seems odd that the cause of action could accrue two and one-half years before the injury even occurs. How could the law support such an incongruous result? The answer is that it does not. Such a result would be contrary to well-established Wisconsin law.
¶ 70. For years, Wisconsin cases have repeatedly held that "the date of the negligent act and the date of the injury in medical malpractice cases are not always one and the same." Fojut, 212 Wis. 2d at 830; see also Paul v. Skemp, 2001 WI 42, ¶ 20, 242 Wis. 2d 507, 625 N.W.2d 860 ("The plain language of Wis. Stat. § 893.55(l)(a) indicates that it is not the negligence, but the injury resulting from the negligent act or omission which initiates the limitations period.").
¶ 71. There are strong policy reasons for keeping negligence and injury analytically separate. "If a negligent act or omission... triggered the limitations pe*590riod. . . potential claimants who have not yet been injured would be seeking relief for damages that may never occur." Paul, 242 Wis. 2d 507, ¶ 42.
¶ 72. What is particularly troublesome is that all of the uncertainty engendered by the majority's analysis is unnecessary. The majority has no need to determine that "the presence of an infection-causing sponge" is an injury.
¶ 73. The estate's claim for medical malpractice was not filed until August 9, 2006, three years and one day after the circuit court determined that the action accrued. Even if Robert's injury occurred as late as August 8, 2003, the estate's medical malpractice claim accrued more than three years before the claim was filed. Thus, I agree that the claim is barred by the statute of limitations.
II
¶ 74. I also write separately to comment on the majority's categorization of a previous court of appeals statement as "dicta." See majority op., ¶ 39, ¶ 40 n.8 (discussing Miller v. Luther, 170 Wis. 2d 429, 489 N.W.2d 651 (Ct. App. 1992)).
¶ 75. The contours of a jurisprudence are shaped in part by how dicta is defined and applied. Yet, it has recently been observed that because judges often select definitions as needed for the resolution of a particular case, the definitions of dicta vary from jurisdiction to jurisdiction and across courts: "Through a loose set of practices that vary considerably across jurisdictions, and, perhaps more problematically, across courts and cases, judges, entirely on their own, define such terms as needed to assist in the task of resolving particular cases." Michael Abramowicz & Maxwell Stearns, Defining Dicta, 57 Stan. L. Rev. 953, 958 (2005).
*591¶ 76. The problem with the majority opinion lies not in the varying and often inconsistent definitions of dicta across multiple jurisdictions or courts. Rather, the problem is that the majority fails to elucidate a clear and consistent definition of dicta within this one opinion. Within the same opinion, it provides four different tests for determining whether a statement is dicta.
¶ 77. It first describes dicta as a statement that is "not necessary to the theory on which the court... decided the case." Majority op., ¶ 39. Next, it employs a three-part test defining dicta as a statement "which [1] extends beyond the facts in the case and [2] is broader than necessary and [3] not essential to the determination of the issues before it." Id. Then, focusing on the "relevan[cy]" and the "reasoning," it defines dicta as a statement that is "not relevant to the reasoning of the court in deciding the case." Id., ¶ 40. Finally, it focuses on the necessity and the holding and defines dicta as a statement that is "not necessary to the holding." Id., ¶ 40 n.8.
¶ 78. The majority fails to acknowledge, as we have previously explained, that in Wisconsin there are two lines of cases defining dicta. State v. Leitner, 2002 WI 77, ¶ 22 n.16, 253 Wis. 2d 449, 646 N.W.2d 341. Under one line of cases, a court's discussion of a question "germane to . . . the controversy" is not dicta:
It is deemed the doctrine of the cases is that when a court of last resort intentionally takes up, discusses, and decides a question germane to, though not necessarily decisive of, the controversy, such decision is not a dictum but is a judicial act of the court which it will thereafter recognize as a binding decision.
State v. Picotte, 2003 WI 42, ¶ 61, 261 Wis. 2d 249, 661 N.W.2d 381 (quoting Chase v. Am. Cartage Co., 176 Wis. *592235, 238, 186 N.W. 598 (1922)); see also State v. Kruse, 101 Wis. 2d 387, 392, 305 N.W.2d 85 (1981) ("While the statement in [a prior case] was not decisive to the primary issue presented, it was plainly germane to that issue and is therefore not dictum.").
¶ 79. The court of appeals has also noted that "[w]hen an appellate court intentionally takes up, discusses and decides a question germane to a controversy, such a decision is not a dictum but is a judicial act of the court which it will thereafter recognize as a binding decision." State v. Holt, 128 Wis. 2d 110, 123, 382 N.W.2d 679 (Ct. App. 1985); see also State v. Sanders, 2007 WI App 174, ¶ 25, 304 Wis. 2d 159, 737 N.W.2d 44; State v. Rushing, 2007 WI App 227, ¶ 12, 305 Wis. 2d 739, 740 N.W.2d 894.
¶ 80. A competing line of cases defines dicta as "a statement or language expressed in a court's opinion which extends beyond the facts in the case and is broader than necessary and not essential to the determination of the issues before it." State v. Sartin, 200 Wis. 2d 47, ¶ 60 n.7, 546 N.W.2d 449 (1996). This is one of the four definitions of dicta used by the majority in this case. See majority op., ¶¶ 39, 40.
¶ 81. Yet, the majority never discusses whether the statement it dismisses as dicta was "germane to the controversy before the court." It never acknowledges the line of cases quoted above. Instead, it simply ignores the question.4
*593¶ 82. How a court defines and applies dicta is important to our common law tradition of fidelity to prior cases. The principle of stare decisis — the obligation to adhere to past opinions — "promotes the evenhanded, predictable, and consistent development of legal principles." Jordan Wilder Connors, Treating Like Subdecisions Alike: The Scope of Stare Decisis as Applied to Judicial Methodology, 108 Colum. L. Rev. 681 (2008). The constraint placed on courts by stare decisis inhibits courts from acting arbitrarily and capriciously.
¶ 83. Yet, loose and unpredictable standards for determining whether a statement is dicta can undermine stare decisis and the principles of judicial restraint. As Judge Brown of the court of appeals has stated, "[t]he term 'dicta'... is often too broadly defined, usually by a lawyer who is searching for a way not to be bound by a prior published decision." Sanders, 304 Wis. 2d 159, ¶ 41 (Brown, J., concurring).
*594¶ 84. Commentators have noted that "[a]s the distinction between holding and dicta becomes increasingly vague, past precedents can be increasingly manipulated. Judges will face greater temptation to cheat... when they can offer some facially plausible argument for disregarding a statement in a prior case." Abramowicz & Stearns, supra, at 1024.
¶ 85. Here, the majority fails to elucidate a clear standard for determining whether a court's statement is dicta. Instead, it employs the term dicta selectively to dismiss a difficult proposition from a prior decision without meaningful analysis. This end run around stare decisis undermines our common law tradition of fidelity to precedent.
¶ 86. For the reasons discussed above, I respectfully concur in part and dissent in part.
¶ 87. I am authorized to state that Chief Justice SHIRLEY S. ABRAHAMSON and Justice N. PATRICK CROOKS join this concurrence/dissent.There are a plethora of medical malpractice sponge cases. Even sponges which reportedly were sterile have been the subject of claims involving infectious injury. See, e.g., Faherty v. Gracias, 874 A.2d 1239, 1247 (Pa. Super. 2005); Hutchins v. Fletcher Allen Health Care, 776 A.2d 376, 377 (Vt. 2001); Fritz v. Horsfall, 163 P.2d 148, 161 (Wash. 1945).
One such case is Seals v. Gosey, 565 So. 2d 1003 (La. App. 1990). There, the court observed that it was the consensus of medical experts that sponges which are initially sterile can be the nesting place for infections to grow, thereby exacerbating an infectious condition. In that instance, the sponge may not be the cause but rather is a cause of the infectious injury. "The medical experts unanimously testified that sterile gauze will not itself cause an infection but that it can be a nesting place for an infection to grow." Id. at 1010.
The majority can only assert that the infection developed "soon" after the July 24 surgery. Majority op., ¶ 3. This fact is consistent with the record, but it does not establish when the sponge became "infection-producing."
Only two and one-half weeks elapsed between the first surgery and Robert's death. At some time during this period, the infection developed. Stating that the infection developed "soon" after the surgery does nothing to pinpoint when in fact the sponge started producing an infection.
See Bowers v. Olch, 260 P.2d 997 (Cal. App. 2d 1953).
The majority never examines whether the Miller court intentionally took up, discussed, and decided the question of when Miller's wife's claim for wrongful death accrued. See majority op., ¶¶ 35-45 (discussing Miller v. Luther, 170 Wis. 2d 429, 489 N.W.2d 651 (Ct. App. 1992)). Miller died on October 22, 1990, and his wife filed a wrongful death action based on medical malpractice one month after his death. Miller, 170 *593Wis. 2d at 434. The circuit court denied the defendants' motion for summary judgment, concluding that the wrongful death action was timely because it was filed within three years of death. Id. at 435.
The court of appeals reversed the circuit court's decision. Id. at 442. It determined that even though the action was not barred by the wrongful death statute of limitations, it could not be maintained for other reasons. Id. The court stated: "As indicated earlier, a wrongful death action accrues at the time of death, and this opinion does not change that rule." Id.
Surely, this proposition was germane to the issue before the court — the statute of limitations for a wrongful death action was the basis of the circuit court's decision. Further, if the claim was barred by the wrongful death statute of limitations, the court would have no need to reach out and decide that the claim could not be maintained on other grounds.